1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
|
/*******************************************************************************
* Copyright (c) 2010-2015, Andras Szabolcs Nagy, Abel Hegedus, Akos Horvath, Zoltan Ujhelyi and Daniel Varro
* This program and the accompanying materials are made available under the
* terms of the Eclipse Public License v. 2.0 which is available at
* http://www.eclipse.org/legal/epl-v20.html.
*
* SPDX-License-Identifier: EPL-2.0
*******************************************************************************/
package org.eclipse.viatra.dse.objectives;
import java.util.ArrayList;
import java.util.Arrays;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import java.util.Random;
import org.eclipse.viatra.query.runtime.matchers.util.Preconditions;
/**
* This class is responsible to compare and sort fitness values. {@link TrajectoryFitness} instances can be added to an
* instance of this class, that it can sort them.
*
* @author András Szabolcs Nagy
*/
public class ObjectiveComparatorHelper {
private IObjective[][] leveledObjectives;
private List<TrajectoryFitness> trajectoryFitnesses = new ArrayList<TrajectoryFitness>();
private Random random = new Random();
private boolean computeCrowdingDistance = false;
public ObjectiveComparatorHelper(IObjective[][] leveledObjectives) {
this.leveledObjectives = leveledObjectives;
}
public void setComputeCrowdingDistance(boolean computeCrowdingDistance) {
this.computeCrowdingDistance = computeCrowdingDistance;
}
/**
* Compares two fitnesses based on hierarchical dominance. Returns -1 if the second parameter {@code o2} is a better
* solution ({@code o2} dominates {@code o1}), 1 if the first parameter {@code o1} is better ({@code o1} dominates
* {@code o2}) and returns 0 if they are non-dominating each other.
*/
public int compare(Fitness o1, Fitness o2) {
levelsLoop: for (int i = 0; i < leveledObjectives.length; i++) {
boolean o1HasBetterFitness = false;
boolean o2HasBetterFitness = false;
for (IObjective objective : leveledObjectives[i]) {
String objectiveName = objective.getName();
int sgn = objective.getComparator().compare(o1.get(objectiveName), o2.get(objectiveName));
if (sgn < 0) {
o2HasBetterFitness = true;
}
if (sgn > 0) {
o1HasBetterFitness = true;
}
if (o1HasBetterFitness && o2HasBetterFitness) {
continue levelsLoop;
}
}
if (o2HasBetterFitness && !o1HasBetterFitness) {
return -1;
} else if (!o2HasBetterFitness && o1HasBetterFitness) {
return 1;
}
}
return 0;
}
/**
* Adds a {@link TrajectoryFitness} to an inner list to compare later.
*
* @param trajectoryFitness
*/
public void addTrajectoryFitness(TrajectoryFitness trajectoryFitness) {
trajectoryFitnesses.add(trajectoryFitness);
}
/**
* Clears the inner {@link TrajectoryFitness} list.
*/
public void clearTrajectoryFitnesses() {
trajectoryFitnesses.clear();
}
/**
* Returns the inner {@link TrajectoryFitness} list.
*/
public List<TrajectoryFitness> getTrajectoryFitnesses() {
return trajectoryFitnesses;
}
/**
* Returns a random {@link TrajectoryFitness} from the pareto front.
*/
public TrajectoryFitness getRandomBest() {
List<TrajectoryFitness> paretoFront = getParetoFront();
int randomIndex = random.nextInt(paretoFront.size());
return paretoFront.get(randomIndex);
}
/**
* Returns the pareto front of the previously added {@link TrajectoryFitness}.
*/
public List<TrajectoryFitness> getParetoFront() {
return getFronts().get(0);
}
/**
* Returns the previously added {@link TrajectoryFitness} instances in fronts.
*/
public List<? extends List<TrajectoryFitness>> getFronts() {
Preconditions.checkArgument(!trajectoryFitnesses.isEmpty(), "No trajectory fitnesses were added.");
List<ArrayList<TrajectoryFitness>> fronts = new ArrayList<ArrayList<TrajectoryFitness>>();
Map<TrajectoryFitness, ArrayList<TrajectoryFitness>> dominatedInstances = new HashMap<TrajectoryFitness, ArrayList<TrajectoryFitness>>();
Map<TrajectoryFitness, Integer> dominatingInstances = new HashMap<TrajectoryFitness, Integer>();
// calculate dominations
for (TrajectoryFitness TrajectoryFitnessP : trajectoryFitnesses) {
dominatedInstances.put(TrajectoryFitnessP, new ArrayList<TrajectoryFitness>());
dominatingInstances.put(TrajectoryFitnessP, 0);
for (TrajectoryFitness TrajectoryFitnessQ : trajectoryFitnesses) {
int dominates = compare(TrajectoryFitnessP.fitness, TrajectoryFitnessQ.fitness);
if (dominates > 0) {
dominatedInstances.get(TrajectoryFitnessP).add(TrajectoryFitnessQ);
} else if (dominates < 0) {
dominatingInstances.put(TrajectoryFitnessP, dominatingInstances.get(TrajectoryFitnessP) + 1);
}
}
if (dominatingInstances.get(TrajectoryFitnessP) == 0) {
// p belongs to the first front
TrajectoryFitnessP.rank = 1;
if (fronts.isEmpty()) {
ArrayList<TrajectoryFitness> firstDominationFront = new ArrayList<TrajectoryFitness>();
firstDominationFront.add(TrajectoryFitnessP);
fronts.add(firstDominationFront);
} else {
List<TrajectoryFitness> firstDominationFront = fronts.get(0);
firstDominationFront.add(TrajectoryFitnessP);
}
}
}
// create fronts
int i = 1;
while (fronts.size() == i) {
ArrayList<TrajectoryFitness> nextDominationFront = new ArrayList<TrajectoryFitness>();
for (TrajectoryFitness TrajectoryFitnessP : fronts.get(i - 1)) {
for (TrajectoryFitness TrajectoryFitnessQ : dominatedInstances.get(TrajectoryFitnessP)) {
dominatingInstances.put(TrajectoryFitnessQ, dominatingInstances.get(TrajectoryFitnessQ) - 1);
if (dominatingInstances.get(TrajectoryFitnessQ) == 0) {
TrajectoryFitnessQ.rank = i + 1;
nextDominationFront.add(TrajectoryFitnessQ);
}
}
}
i++;
if (!nextDominationFront.isEmpty()) {
if (computeCrowdingDistance) {
crowdingDistanceAssignment(nextDominationFront, leveledObjectives);
}
fronts.add(nextDominationFront);
}
}
return fronts;
}
/**
* Executes the crowding distance assignment for the specified front.
*
* @param front
*/
public static void crowdingDistanceAssignment(List<TrajectoryFitness> front, IObjective[][] leveledObjectives) {
for (TrajectoryFitness InstanceData : front) {
// initialize crowding distance
InstanceData.crowdingDistance = 0;
}
for (final IObjective[] objectives : leveledObjectives) {
for (final IObjective objective : objectives) {
final String m = objective.getName();
TrajectoryFitness[] sortedFront = front.toArray(new TrajectoryFitness[0]);
// sort using m-th objective value
Arrays.sort(sortedFront, (o1, o2) -> objective.getComparator().compare(o1.fitness.get(m), o2.fitness.get(m)));
// so that boundary points are always selected
sortedFront[0].crowdingDistance = Double.POSITIVE_INFINITY;
sortedFront[sortedFront.length - 1].crowdingDistance = Double.POSITIVE_INFINITY;
// If minimal and maximal fitness value for this objective are
// equal, then do not change crowding distance
if (sortedFront[0].fitness.get(m) != sortedFront[sortedFront.length - 1].fitness.get(m)) {
for (int i = 1; i < sortedFront.length - 1; i++) {
double newCrowdingDistance = sortedFront[i].crowdingDistance;
newCrowdingDistance += (sortedFront[i + 1].fitness.get(m) - sortedFront[i - 1].fitness.get(m))
/ (sortedFront[sortedFront.length - 1].fitness.get(m) - sortedFront[0].fitness.get(m));
sortedFront[i].crowdingDistance = newCrowdingDistance;
}
}
}
}
}
}
|
